Create compute canada env

[liellnima]

See this previous issue here

Compute Canada / Alliance needs a different procedure to create the environment, since conda cannot be used directly, and we need to use module load for cdo and nco.

Requirements:

• [ ] use module load for cdo and nco
• [ ] add documentation on how to set up your env on compute canada
• [ ] add different subsections in readme: quick setup for local / mila cluster / compute canada
• [ ] add simple test that checks if all dependencies could be imported

Additional information: We agreed on the following set to start the development of this package:

Python:

• 3.11

Additional dependencies:

• cdo
• nco
• xesmf
• esgf-pyclient
• xarray

Poetry dependencies:

• numpy
• netCDF4
• xmip
• tqdm
• pint

[f-PLT]

@liellnima @mjfortier

I have a question on this.

From the documentation, I'm under the impression that on DRAC, they want us to install only from their own wheels (using virtualenv --no-downloads, and pip install --no-index)

Virtualenv is no issue, as we can create the virtualenv and use it with Poetry, and replicating --no-index can be worked around.

However, trying to install the dependencies, I found out that esgf-pyclient is not available in their local wheels (meaning we need to install it from the internet).

So, the actual question, how do you create your environments on DRAC usually? Do you use only the local wheel? Only from internet to get the latest versions? Dare I say a mix of both?